Intangible Assets: How the Interaction of Computers and …ebusiness.mit.edu/erik/itqo final-7-00.pdf · Intangible Assets: How the Interaction of Computers and Organizational Structure

Intangible Assets:

How the Interaction of Computers and Organizational

Structure Affects Stock Market Valuations

Erik Brynjolfsson

MIT Sloan School of Management50 Memorial Drive, Suit E53-313Cambridge, MA 02142-1347 USA

[email protected]

Lorin M. Hitt

University of Pennsylvania, Wharton School1318 Steinberg Hall-Dietrich Hall, 3620 Locust Walk

Philadelphia, PA 19104-6366 [email protected]

Shinkyu Yang

New York University, Stern School44 West 4th St., Suite KMEC 9-76New York, NY 10012-1106 USA

[email protected]

Acknowledgements: We thank Robert Gordon, Zvi Griliches, Bronwyn Hall, RobertHall, Boyan Jovanovich, Jacques Mairesse, Paul Romer, Eli Snir and the participants atthe Workshop on Information Systems and Economics, the MIT Center for CoordinationScience, National Bureau of Economics Research, the Wharton School, and the AEAannual meetings for valuable comments on earlier versions of this paper. The NationalScience Foundation (IIS-9733877), the Organization for Economic Cooperation andDevelopment, the Stanford Computer Industry Project, and IBM Research providedgenerous funding.

Intangible Assets:

How the Interaction of Computers and Organizational Structure

Affects Stock Market Valuations

This paper investigates the proposition that the widespread use of information technology

has increased investment in intangible organizational assets. Using firm-level data, we

find that each dollar of installed computer capital in a firm is associated with at least five

dollars of market value, after controlling for other assets. We interpret this value as

revealing the existence of a large stock of intangible assets that are complementary with

computer investment. Using data on organizational practices at each firm, we identify a

specific cluster of practices that appear to represent at least some portion of these

intangible assets. Not only are these practices correlated with computer investments, but

firms that combine higher computer investments with these organizational characteristics

have disproportionate increases in their market valuations. We conclude that investors

believe that the contribution of computers is increased when they are combined with

certain intangible assets, specifically including the cluster of organizational changes that

we have identified.

Intangible Assets Page 1

1. Introduction

In developed economies, production not only requires the traditional factors such as

capital and labor, but skills, organizational structures, know-how, information, and other

factors that are collectively referred to as “intangible assets.” Detailed investigation of

these types of assets has found that they are often large and have substantial productivity

benefits. For example, Jorgenson and Fraumeni (1995) found that the stock of “human

capital” in the U.S. economy dwarfs the stock of physical capital and has grown

substantially over time; B. Hall (1993a), Griliches (1981), and Lev and Sougiannis

(1996) have found that the R&D assets bring benefits in the form of positive marginal

product and market valuation. Results from analyses of “Tobin’s q” have shown that the

stock market valuation of firms has increasingly diverged from their measured book

value (Chan, Lakonishok and Sougiannis, 1999; R. Hall, 1999).

One possible explanation for the recent increase in Tobin’s q increasing importance of

intangible capital is the growing use of information technology and the associated

investments in intangible assets (R. Hall, 1999; Brynjolfsson and Yang, 1999). While

early applications of computers were primarily directed at factor substitution (particularly

of low-skill clerical workers), modern uses of computers and flexible manufacturing

technologies have both enabled and necessitated substantial organizational redesign

(Berman, Bound and Griliches, 1994; Brynjolfsson and Hitt, 1998; Brynjolfsson,

Renshaw and van Alstyne, 1997; Hammer, 1990; Black & Lynch, 1999; Orlikowski,

1992) and changes in the skill mix of employees (Autor, Katz and Krueger, 1998;

Bresnahan, Brynjolfsson and Hitt, 1999). Collectively, these papers argue for a


complementarity between computer investment and organizational investment, and

specifically a relationship between information technology use and increased demand for

skilled workers, greater decentralization of decision rights, and team-oriented production.

Moreover, case studies suggest that these complementary investments are large. For

example, less than 20% of the typical $20 million installation cost of the SAP R/3 system

(a common large-scale package designed to integrate different organizational processes)

is for hardware and software; the vast majority of the investment is for consultants to

customize the software, to redesign organizational processes, and to train the staff in the

use of the new system (Gormely, et. al., 1999).

In this paper, we analytically explore the hypothesis that new intangible organizational

assets complement information technology capital just as factory redesign complemented

the adoption of electric motors (David, 1990) and memos and filing systems

complemented the printing press. To realize the potential benefits of computerization,

investments in additional "assets" like new organizational processes and structures,

worker knowledge and redesigned monitoring, reporting and incentive systems may be

needed.

The presence of intangible assets can be observed in two ways. First, the resulting effect

on the firm’s market valuation should be measurable. While effects on productivity or

other measures of economic output may be spread over many years, the financial

markets, which seek to assess the discounted value of future revenues, provide an

immediate indicator of whether these investments generate value for the firm's owners.


In particular, the market value of a firm which has leveraged computer assets with

organizational investments should be greater than that of a similar firm which has not

incurred these investments. A computer that is integrated with complementary

organizational assets should be significantly more valuable to a business than a computer

in a box on the loading dock.

Second, some of the specific changes that firms make may be directly observable. In

particular, numerous authors have suggested that information technology (IT) is likely to

be associated with organizational changes such as greater demand for worker skills and

increased levels of employee decision-making authority (Applegate, Cash and Mills,

1988; Bresnahan, Brynjolfsson and Hitt, 1999; George and King, 1991; Mendelson and

Pillai, 1999; Sauer and Yetton, 1997). If these practices represent the types of

organizational assets we described earlier, then we would expect that the value of IT

would be greater in organizations that also adopt these work practices.

This complementarities argument leads to four implications which are testable in

empirical data. First, each dollar of installed computer capital should be correlated with

more than one dollar of market value, after controlling for other measured assets.

Second, investments in computers should be correlated with increased investments in

certain organizational practices. Third, if these practices represent part of the productive

assets of a firm, they should also be associated with increases in market value. Finally,

firms that combine these specific organizational practices with investments in computer


capital should have a higher market value than those that adopt these same practices in

isolation.

Using data on 1031 large firms over 8 years (1987-1994), we find evidence in support of

all four implications of our argument:

1. Each dollar invested in computers is associated with an increase in firm market

valuation of $5 to $20 (depending on the assumptions of the estimation models),

compared with an increase of about $1 per dollar of investment in other assets.

2. Firms that are high IT users are also more likely to adopt work practices that involve

a cluster of organizational characteristics, including greater use of teams, broader

decision-making authority, and increased worker training.

3. This cluster of organizational characteristics increases a firm's market valuation

beyond what can be accounted for by tangible assets.

4. In firms that use these organizational characteristics, the computer assets have a

disproportionately higher market valuation.

Our results are robust to a variety of alternative estimating techniques. The results are not

consistent with the alternative hypothesis that differences in the speed of adjustment to

unexpected shocks biases the results. Similarly, they are not driven entirely by a general

capital-skill complementarity; these results appear to be unique to IT capital and do not


appear for ordinary capital. Because our sample is predominantly large, established

firms rather than new high technology entrants, and the time period of our data predates

the large increase in the value of technology stocks in the 1990s, our results are not likely

to be affected by the possible existence of a “high-tech stock bubble”. However, they

are consistent with earlier case-based research as well as recent econometric work using

production functions which suggest an important role for IT-enabled organizational

changes in increasing productivity and the value of firms. Taken together, these results

lend quantitative support to the idea that IT is most valuable when coupled with

complementary changes in organizational design.

In Section 2, we present a sketch of the theoretical model and the data; in Section 3 we

present our statistical results, and we conclude with a summary and discussion in Section

4.

2. Econometric Model and Data

2.1 Derivation of Model for Stock Market Valuations

In this subsection, we sketch the derivation of the stock market valuation model. The

basic structure of the model follows the literature on the valuation of capital goods that

relates the market value of a firm to the capital goods a firm owns (Hayashi, 1982;

Wildasin, 1984; Hayashi and Inoue, 1991, Brynjolfsson and Yang, 1997). This literature

is often referred to as the "Tobin's q" literature after the pioneering work by James Tobin


(1969) in describing the relationship between firm value and capital investment. This

framework has been empirically adapted and applied to the valuation of R&D by

Griliches (1981), Griliches and Cockburn (1988), and B. Hall (1993a,b; 1999) and the

stock market impact of diversification (Montgomery and Wernerfelt, 1988) using firm-

level data.

The empirical use of Tobin’s q to capture intangible organizational assets has been

proposed by other authors. R. Hall (1999a) states in his discussion of his quantity

revelation theorem, “the value of corporate securities, interpreted as a measure of the

quantity of capital, behaves reasonably” and the firm’s intangible assets are “technology,

organization, business practices, and other produced elements of the successful modern

corporation.” R. Hall (1999b) also discusses the analogy between a flow of investment in

reorganization and a flow of investment in physical capital. Our paper is closely related

to Brynjolfsson and Yang (1999) who found evidence of high q values for IT, but did not

explicitly link them to organizational investments.

We assume that firms face a dynamic optimization problem in which managers make

capital investments (I) in several different asset types and expenditures in variable costs

(N) with the goal of maximizing the market value of the firm V. In turn, V is equal to the

present value of all future profits with a discount function u(t). The accumulation of

capital investment, less depreciation (δ), produces a vector of capital stock (K, which

includes different components of capital Kj, j=1…J where the j’s are physical capital,

computers, etc.). The capital stock along with variable inputs is used to produce output


via a production function (F). We depart from most of the traditional literature1 by

assuming that there is some additional cost of making a capital investment which

represents an "organizational adjustment cost,” (Γ(I,K,t)). These organizational costs

represent the amount of output lost while integrating additional capital into the firm. This

yields the following program:

(1) Maximize V t u t dtI N,

( ) ( ) ( ) 00

=∞zπ

(2) where π( ) ( ( , , ) ( , , ))t F K N t I K t N I= − − −Γ

(3) and the following holds: dKdt

I Kjj j

j

J

= −=

∑δ1

, for all j =1, . . . , J.

One can solve for the market value of the firm that results from this optimization problem

with additional assumptions on the structure of F(·) and Γ(·).2 If there are no

organizational adjustment costs are needed to make capital assets fully productive

(Γ(I,K,t) = 0), then buying a firm is no different from buying a collection of separate

assets. Thus, the market value of a firm is simply equal to the current stock of capital

assets:

1 See Yoshikawa (1980) and Wildasin (1984) for models where Tobin’s q is affected byadjustment costs.2 We assume that F(K,N) and Γ(I,K) are homogeneous functions of degree 1 over K,N, and I(constant returns to scale) and are twice differentiable. We further assume that Γ(I,K) isincreasing and convex in investment, with no fixed costs (Γ(0,K)=0), and is non-negativeeverywhere. A detailed derivation of this estimating equation appears in Brynjolfsson and Yang(1999).


(4) V K jj

J

==

∑1

However, if organizational adjustment costs are required to make full use of capital, then

the value of an ongoing firm may exceed the value of its separate capital assets. The

higher value can be thought of representing the additional "intangible assets" created

when each capital asset is integrated into the firm. In this case, the value of the firm is

the sum of capital assets, but weighted by the size of the organizational adjustment costs,

λ:

(5) V Kj jj

J

==

∑λ1

.

For example, if there are two types of capital, computers (Kc) and other capital (Kp), then

(λc –1) would represent the difference in value between computer capital which is fully

integrated into the firm and computers which are available on the open market, and (λp –

1) would be the corresponding value for other types of capital. We can then calculate

the size of the complementary organizational investments by comparing how much the

market values a capital asset which is part of an operating firm as compared to the same

asset sold separately.


In addition to the capitalized adjustment costs, there may also be various intangible assets

correlated with each of the Kj. When (νj -1)K j is the other intangible assets correlated

with Kj, then the market value equation (5) becomes:

(6) V Kj j jj

J

==

∑ν λ1

.

2.2 Econometric Issues of Market Valuation

To empirically estimate the relationship in equation (6) it is necessary to specify the

different types of capital assets that we will consider as well as a set of additional control

variables (X) to account for sample heterogeneity. In addition, we will sometimes

include a fixed effect, α, to capture differences across firms that are constant over time,

thereby further controlling for firm heterogeneity. Including an error term, ε, we have our

general estimation equation:

(7) V K Xit i j j it it itj

J

= + + +=

∑α λ γ ε,1

Here, i, t, and j are indices of firms, time, and different capital goods, respectively. The

coefficients to be estimated are (vectors) α, λ, and γ.


Extending the prior literature on estimates of Tobin's q, we divide assets into three

categories: computers, physical assets (property, plant and equipment), and other balance

sheet assets (receivables, inventories, goodwill, cash, and other assets). For the other

control variables (X) we will use the ratio of R&D capital to assets, and the ratio of

advertising expense to assets industry dummies (usually at the SIC 2-digit level), and

year dummies.3 This yields our base estimating equation which we will later extend to

explicitly include certain organizational investments:

(8) V K K K controlsit i c c it p p it o o it it= + + + + +α λ λ λ ε, , ,

Here Kc, Kp, and Ko represent computer capital, physical capital, and other balance sheet

assets, respectively. This methodology can be considered an example of hedonic

regression, which estimates the market shadow "price" for various assets using cross-

sectional and time series variations in their quantities and in the market value of the firm

(B. Hall, 1999).

Because firm sizes vary substantially in our sample and our model is implemented in

levels rather than logarithms, we anticipate significant size-related heteroskedasticity.

We address this problem by using generalized (weighted) least squares (GLS) as well as

3 Advertising and R&D are other types of nonstandard assets that have been considered in priorwork. Because no capitalized value of these "assets" is reported, we simply include them as ratiosin the reported regression. This can be though of treating current spending on these assets as anoisy indicator of their capital stock values (B. Hall, 1993a,b; see also Brynjolfsson and Yang,1999 for a more detailed analysis of these assets in this context). Finally, we add additionalcontrol variables for industry to reduce sample heterogeneity, and time to control for generaleconomic trends in stock market valuation.


robust regression techniques (least absolute deviation - LAD) which are less sensitive to

outliers of all sorts. We also address explicitly several sources of potential specification

error in our analysis. One concern is that computers may be disproportionately correlated

with other unobserved, but valuable, firm characteristics. Therefore, the measured

marginal value of computers (essentially a hedonic price) will include not only the

physical computers, but also the intangible assets that were excluded from the equation

but are correlated with computers. This is not problematic in a larger sense, as it is

perfectly consistent with the central story of this paper. However, we can separate out

these effects in two ways. First, we can perform a fixed-effects ("within") regression that

removes all time-invariant firm characteristics, and thus facilitates one estimate of the

value of complementary organizational assets. Alternatively, we can explicitly include

some measurable components of organization in the equation as an additional variable

and measure its direct value and its interaction effect with computers.

Another potential difficulty is the endogeneity of computer investment. While our model

seeks to measure whether changes in the value of a firm's capital assets affect its stock

market value, it may also be the case that unexpected increases in stock market valuations

lead firms to make greater investments in capital assets. For example, an unexpected

shock to demand may increase investors’ long-run expectations of profitability

(increasing market value) and encourage the firm to expand production capacity. This is

essentially the behavioral model that underlies the q-theory of investment (Tobin, 1969).

If both the real investment and the financial markets adjust instantaneously to such

shocks, then the coefficient values we observe can be interpreted as the correct "prices"


for each asset. Even short adjustment lags are not necessarily a problem. Over time,

efficient markets will cause these types of shocks to average out; thus they may not be

substantial as long as the adjustment period is relatively short (R. Hall, 1999). However,

endogeneity could be problematic if the lags are fairly long relative to the estimation

period. One approach to this problem is to lag the capital stock variables by one period.

Few managers can forecast the unexpected components of stock returns better than the

market and keep those forecasts a secret. Therefore, previous period investment can be

considered predetermined relative to future market value shocks.

An additional difficulty arises if capital stocks do not adjust instantaneously. For

example, suppose a demand shock arrives that requires increased use of IT and increases

market value, but only some of the IT can be installed in time. This will tend to

understate the current capital requirements and could lead to an overestimation of the

market value of capital. However, the market value estimates that results from such

shocks simply represent the capitalized value of future investments made possible by the

firm's current investment position. In essence, they are the quasi-rents of having an

installed base of computers, capital, and intangible assets. Under this interpretation, the

per-unit value of IT may be larger than its long-run value, but still accurately reflects its

current value. Future investments will in IT will not increase the market value

incrementally, but will be essential to maintain the market value level already achieved.

A variant of this issue arises when the firm must adjust to more than one type of capital,

and these adjustments occur at different speeds. For example, IT may adjust very


quickly to unanticipated shocks, while ordinary capital might require several years to

adjust. In this case, especially for analyses conducted on first differences, more of the

variation in market value will be explained by the newly-acquired computers since the IT

appears to have a larger change than capital. Fortunately, this bias problem is reduced

when the analysis is conducted over longer time horizons. In a one-year difference, small

differences in adjustment rate will lead to large differences in capital stocks, but over 3 to

5 years all factors will be proportionally closer to their equilibrium value. We can

therefore construct a test to determine whether varying adjustment rates lead to biases:

conduct the analysis using different difference lengths. If computers are erroneously

receiving credit for some of the benefits that should be attributed to slower-adjusting

capital, the computer coefficient should fall as longer time periods are considered.

Conversely, if the coefficients on computers rise as the time period considered lengthens

(above and beyond any effects due to measurement error4), then this provides evidence

that variation in adjustment rates is not biasing the results upward.

2.3 Data Sources and Construction

The data set used for this analysis is a panel of computer capital and stock market

valuation data for approximately 1000 firms over the 1987-1994 time period,

matched to a cross-sectional survey of organizational practices conducted in 1995

and 1996. A brief description of each data source follows, with additional detail

in the Appendix.

4 Longer differences decrease bias when only a single variable is measured with error (or all thedependent variables are orthogonal). See Griliches and Hausman (1986).


Computer Technology: The measures of computer use were derived from the

Computer Intelligence Infocorp installation database that details IT spending by

site for Fortune 1000 companies. Approximately 25,000 sites were aggregated to

form the measures for the 1000 companies that represent the total population in

any given year. This database is compiled from telephone surveys that detail the

ownership of computer equipment and related products. Most sites are updated at

least annually with more frequent sampling for larger sites. The year-end state of

the database from 1987 to 1994 was used for the computer measures. From this

data we obtained the total capital stock of computers (central processors, personal

computers, and peripherals). The IT data do not include all types of information

processing or communication equipment and are likely to miss some portion of

computer equipment that is purchased by individuals or departments without the

knowledge of information systems personnel. 5

Organizational Practices: The organizational practices data in this analysis come

from a series of surveys of large firms. These surveys adapted questions from

prior surveys on human resource practices and workplace transformation

(Huselid, 1995; Ichniowski, Shaw and Prunnushi, 1997; Osterman, 1994). The

questions address the allocation of various types of decision-making authority, the

5 Another potential source of error in this regard is the outsourcing of computer facilities.Fortunately, to the extent that the computers reside on the client site, they will still be properlycounted by CII’s census. To the extent that these facilities are located at a third-party site, theywill not be properly counted. However, despite all these potential limitations these data arebroadly consistent with other survey work on a more limited sample from International Data


use of self-managing teams, the breadth of job responsibilities and other

miscellaneous characteristics of the workplace (further detail appears in the

results section). Organizational data were collected in three waves at the end of

1995 and early 1996, covering most of the Fortune 1000. This yielded a single

cross section of 416 firms with a survey response rate of 49.7%. We detected no

significant pattern of response bias when compared with the population of firms

in the Fortune 1000.

Compustat . Compustat data were used to construct stock market valuation metrics

and provide additional firm information not covered by other sources. Measures

were created for: total market value (market value of equity plus book value of

debt); property, plant and equipment (PP&E); other assets; R&D assets; and

advertising expense.

Overall, the full data set includes 4592 observations over 8 years for market value and

computer capital stock. By matching these data to the organizational practices surveys,

we obtained complete organizational and market value data for 250 firms for a total of

1707 observations.

3. Results

In this section, we report on the regression and correlation analyses performed to test the

four implications of our complementarities argument outlined in the introduction. First,

Group that measured the stock of IT capital at the firm-level as well as capital flow tables from


we explored the basic relationship between IT and stock market value for our full sample

of firms. We then examined the relationship between computer capital and the adoption

of specific organizational practices using correlation analyses, and constructed a single

variable, ORG, which captured most of the relevant variation in organization across

firms. Third, we investigated the effect of ORG on the firm market value. Finally, we

studied how the combination of ORG and computers affects market value. We also

performed a number of robustness checks of our analysis and considered alternative

hypotheses, as we report in each section.

3.1 Computers and market value 6

3.1.1. Basic Findings for Computers and Market Value

The regression analyses (Equation (8)) for estimating the effect of computers on market

value are shown in Table 1a. As shown in the first column (ordinary least squares

regression), we found that each dollar of installed property, plant and equipment (PP&E)

is valued at about one dollar, which is what theory would predict if these assets are in

equilibrium. A dollar of other assets, which includes accounts receivable, inventories and

liquid assets, is valued at about $0.7; apparently stockholders do not believe they will

receive the full face value of these assets, on average. Strikingly, each dollar of

computer capital is associated with nearly $17 of market value. This implies that the

stock market imputes an average of $16 of intangible assets to a firm for every $1 of

the U.S. Bureau of Economic Analysis (Brynjolfsson and Hitt, 2000).6 This subsection parallels work discussed in Brynjolfsson and Yang (1999).


computer capital. All capital stock variables are significantly different from zero, and the

high R2 (>85%) suggests that we can explain much of the variation in market value

across firms and time with our model. 7

To probe this result further we investigated how much the correlation between market

value and computer investment was driven by variation across firms (a "between"

regression) and variation for the same firm over time (a "firm effects" regression). We

found that both sources of variation were important but that the effect due to variation

between firms was larger. The "between" regression implies a market value of computer

capital of nearly $20. For the firm effects regression, this value is $5 (but still strongly

significant). The firm effects regression can be interpreted as removing all the effects that

are unique to a particular firm but constant over time. This suggests that firm-specific

factors account for a substantial amount of the “excess market valuation” of computers.8

In Figure 1 and Figure 2, we present the relative size of computer coefficients and those

of other assets.

In Table 1b, we examine the robustness of the base results to variations in econometric

methods. For this analysis we restricted the sample to a balanced panel9 to get maximum

data consistency and applied different regression techniques: generalized least squares

7 Among control variables, R&D to asset ratios and advertisement to asset ratios are not alwayssignificant. Firm effects, industry effects, and year effects as separate groups are always stronglysignificant.8 In other words, the difference in intangible assets between highly computerized firms and lesscomputerized firms is greater, on average, than the difference within any single firm over time.9 In other words, we excluded all firms which were missing any data in any year.


(GLS) and least absolute deviation (LAD) regression10 to control for heteroskedasticity

and outliers. Overall, the basic results are consistent whether we use balanced or

unbalanced panels and GLS or LAD in both the between and fixed effects regressions.11

3.1.2 Leads and Lags of Computers and Market Value

The estimation of Equation (8) can be interpreted as a hedonic regression, where the

value of a firm can be decomposed to the values of its component assets (Hall, 1993b;

1999). This interpretation is valid when shocks in stock market value do not affect the

investment behavior of firms. However, according to the standard q theory of investment,

if the desired level of capital is influenced by market value shocks, then the estimates of

λ may be biased. In particular, when there exist non-zero time lags for adjustment and the

lag differs for each type of capital, the biases are likely to be larger.

In Table 1c we explore this possibility empirically, adapting a framework based on the

work of Granger (1969). For each regression shown in the table, we regressed the current

period value of a measure (either IT or market value) on the lags of the other measure.

As shown in the first and second columns of the table, the changes in current and lagged

market values did not affect current computer investments. These columns represent

regressions of lagged computers and market value on current computer quantity. In our

10 LAD regression minimizes the absolute value of the deviation of the actual and fitted values, asopposed to the square of the difference as is done for OLS. Standard errors for the LADestimates are calculated using bootstrapping techniques with 100 repetitions to obtain theempirical distribution of the coefficient estimates.11 While a plot of regression residuals (not shown) suggests strong size-based heteroskedasticity,the results changed very little with alternative estimation methods.


data, we did not observe that investments responded significantly to changes in market

value. This result is consistent with studies that report that investment shows limited

response to changes in q value (Abel and Blanchard, 1986; Jorgenson, 1986; Schaller,

1990). 12

In the opposite direction, we found a substantial relationship between past computer

investment and the current increase in market value. As shown in the second column of

Table 1c, investments in computers Granger cause increases in market value in the

“simple causation model” (Granger, 1969), and the coefficients of lagged computer

investments are large. In the “instantaneous causality model”, where current computer

quantity is included in addition to the first three lags (Column 3), current computer

investments also lead to increases in the market value. Given that past investments in

computers are associated with current increases in market value but not vice versa, it is

harder to sustain an argument that the strong correlation between computers and market

value can be attributed to reverse causation. 13 A more plausible interpretation of these

results is that firms continue to build follow-up intangible assets after they invest in

computers, and moreover, that the follow-up investments are large. This explanation is

consistent with a plethora of case evidence (See Brynjolfsson and Hitt, 2000 for a

review).

12 Able and Blanchard (1986) and Schaller (1990) discuss the issue in some detail and suggestalternative explanations.13 If one wishes to maintain the hypothesis that future stock market increases “cause” pastcomputer investments, one could tell a story in which managers have private foreknowledge ofhigher market value in the future and therefore invest in computers but do not take any actions(such as trading in their firms’ securities) which would reveal their private information.Econometrics alone cannot rule out such stories, though they may require increasinglyimplausible assumptions.


These results can also be reconciled with the empirical evidence on the q-theory of

investment. If high q values mainly indicate the presence of intangible assets, rather than

capital-using market opportunities, we would not necessarily expect a firm to accumulate

additional tangible capital when its q value is high. This is consistent with the low

observed correlation between q and capital investment. It is also consistent with the

observation that economy-wide q values have increased in recent years concurrently with

large investments in information technology.

3.1.3 Varying Difference Lengths

If the adjustment speed for investments in computers is higher than the adjustment speed

for investments in other types of capital goods, the coefficient on computers may be

biased upward when unexpected market value shocks occur.14

One way to address this problem is to perform the analysis varying the difference length.

For example, if shocks are relatively infrequent and one type of capital adjusts within one

year while another takes two years, then the variation in adjustment speeds may

significantly influence the coefficient estimates for short differences, but become

unimportant when time horizons over two years are considered. The between regressions

represent the limit of this process, essentially corresponding to a difference with infinite

length. If the computer coefficient falls relative to the capital coefficient over longer

14 This is especially true when the measures of capital stock variables are slow to detect changes.For example, in factors such as ordinary capital, the change in capital stock may be difficult todetect relative to the noise level, while large changes in the IT capital figures will substantially


difference periods, this suggests that some of the effects of capital on the market value

have been erroneously attributed to IT in the short difference analyses. If the coefficient

rises substantially, it suggests that any upward bias from relative adjustment speeds is not

significant. The results of this analysis are shown in Table 1d.

We find that as we move from one-year differences to seven-year differences, the

coefficient on IT generally rises from $3 to about $8, with a dip at fourth-year

differences. The capital coefficient rises slightly from one-year differences to three-year

differences and then stays level at approximately $1.2. This substantial rise in the IT

coefficient, over a factor of 2, suggests that differences in adjustment speed do not lead

us to overestimate the contribution that computers make to market value in our basic

specification.

Another way to examine the robustness of the results is to examine year-by-year cross

sections of the results. If the results are biased upward by short run shocks then some

years will have disproportionately high values while others will be close to their

equilibrium value. In Table 1e we present regressions breaking our 8-year time period

into four two-year intervals. Although there is substantial year to year variation in the

coefficients, there is no particular time trend and in all four subsets the point estimate for

the IT capital coefficient exceeds 11. This again suggests that while there is some time

series variation, short run shocks do not appear to explain the high market valuation of IT

capital.

exceed the noise level. Since bias due to errors in variables is proportional to the “signal-to-noise” ratio, this could lead to upward bias for IT.


Our results favor the argument that the high market valuation of computers reflects the

presence of complementary intangible assets rather than alternative explanations such as

stock price shocks, unobserved heterogeneity, and various forms of adjustment lags.

These assets dwarf the size of actual computer investment and appear to be unique to

computers – no similar effect is found for other types of capital. Moreover, the difference

between the fixed-effects and between regressions suggests that much of these intangible

assets are unique to particular firms. In the following section, we explicitly analyze one

component of these intangible assets, and explore its effect on market value both directly

and through complementarities with computers.

3.2 Basic findings regarding role of organizational structure 15

In this section, we report on the correlations between computers and various measures of

internal organization. All correlations use Spearman rank order correlations 16 between

various measures of computers and the organizational variables, controlling for firm size

(employment), production worker occupation, and industry. 17 We used three different

measures of IT , including the total value of IT installed base (ITCAP), total central

15 These results build on earlier work reported in Hitt and Brynjolfsson (1997) and Bresnahan,Brynjolfsson and Hitt (1999).16 Results are similar when probit or ordered probit regression is used. We report Spearman rankorder correlations because they are easier to interpret given the non-metric nature of most of ourwork system variables.17 Included are separate controls for mining/construction, high technology manufacturing(instruments, transportation, electronics, computers), process manufacturing (paper, chemicals,petroleum), other non-durable manufacturing, other durable manufacturing, transport, utilities,trade, finance, and services.


processing power18 in millions of instructions per second (MIPS), and number of

personal computers (TOTPC). We used multiple measures because they capture slightly

different aspects of computerization (for example, MIPS measures centralized

computing, while TOTPC measures decentralized computing).

In Table 2a we present correlations between multiple measures of IT and four dimensions

of organizational design: use of teams and related incentives, individual decision-making

authority, investments in skills and education, and team-based incentives. These types of

practices have been linked to IT investment in previous theoretical and empirical work

(see Brynjolfsson and Hitt, 1997; Bresnahan, Brynjolfsson and Hitt, 1999 and

Bresnahan, 1997 – a survey of related work appears in Brynjolfsson and Hitt, 2000).

Consistent with our argument that IT and organization are complementary, we confirm

that across multiple measures of IT and multiple measures of organization, firms that use

more IT differ statistically from other firms: they tend to use more teams, have broader

job responsibilities, and allocate greater authority to their workers, even after controlling

for firm size and industry.

In addition to being correlated with IT, these practices are all correlated with each other.

Following Hitt and Brynjolfsson (1997) we constructed a composite variable (ORG) as

the standardized (mean 0, variance 1) sum of the standardized individual work practice

variables. This allowed us to capture an organization's overall tendency to use this

collection of work practices in a single construct, which we can then use for further

analysis. A principal components analysis, Table 2b, showed that all components of this

18 Total central processing power does not include the processing power of personal computers.


variable load highly on a single factor (which explains approximately 35% of the

variance of these measures), and a scree plot (not shown) suggests that this is the only

non-noise factor. The composite variable, ORG, is highly correlated with

computerization, consistent with our earlier arguments. In the remaining section of the

results, we will explore the influence that this cluster of practices has on both the market

value of the firm and the market value of computer capital.

3.3 The Effect of Organizational Structure on Market Value

3.3.1. Organization variable in the market value equation

In this section, we report how we modified our base estimating equation to include

measures of organizational assets. We then investigated the direct relationship of these

measures on market value as well as their effect on the market value of computers

through interaction terms. This yielded the following estimating equation:

(8) V K K K ORG ORG K controlsit i c c it p p it o o it i c it it= + + + + + ⋅ + +α λ λ λ ω ω ε, , , ,1 2

A test of our argument that organizational investments can be treated as intangible assets

is whether the ORG has a positive contribution to market value.


To test our argument that there are positive synergies between IT and organizational

investments we examined whether IT is more valuable in high ORG firms; that is, we

tested the null hypothesis, ω2 = 0 against ω2 ≠ 0.

We examined several market value equations that also include the ORG variable as a

measure of organizational capital. The first three columns of Table 3a report the same

analysis of market valuation of computers for the subsample matching panel data on IT

and other factors and with a cross section of organizational data at the end of the sample

period. The coefficients are qualitatively consistent with the results from the larger

sample shown in Tables 1a and 1b.

When we simply added the ORG variable to the baseline market value equation, we

found that it had a large and statistically significant contribution, as shown in Table 3a,

columns 4 and 5. Firms that are one standard deviation above the mean in ORG have a

market value that is about $500 million higher than the mean, ceteris paribus. Evaluated

at the mean, one standard deviation of the ORG variable corresponds to an 8% increase in

market value.19 Thus, investors appear to treat organizational capital much like more

tangible types of capital by recognizing its contribution to the market value of a firm.

3.3.2. Interaction between organization and computers

19 Results from between regression and pooled regression are essentially similar. The fixed effectspecification is omitted since it is not meaningful to estimate the coefficient of a time-invariantvariable.


In Table 3b, we present the results when we included both ORG and its interaction with

computer capital in the regression. The magnitude of the interaction term between IT

and ORG is about 6 in the pooled estimation, suggesting large complementarities

between computers and organizational structure. In fact, it suggests that each dollar of

computer capital is associated with an increase in market value of an additional six

dollars in firms that are one standard deviation above the average in ORG.

One possible explanation of these results is that ORG makes all types of capital more

valuable and since capital investments tend to be correlated with each other, we are

erroneously attributing this all to computers. When we included additional interaction

terms between ORG and other types of capital (columns 2 and 4 of Table 3b), we found

that this relationship is unique to computers: the coefficients on the added interaction

terms were not significant and the other coefficients changed very little. This indicates

that ORG is an intangible asset that has a particularly strongly effect on the market value

of IT.

Since ORG is measured once per firm at the end of the sample period, we could not apply

a fixed-effect model to estimate its coefficient. However, since computers do vary over

time, so does their interaction with ORG. This enables us to estimate firm effects. The

results (shown in last two columns of Table 3b) provide evidence of an interaction

between ORG and IT even in the firm-effects analysis. The coefficient was reduced

although still significant. When we also removed the direct computer effect (which is

highly collinear with the interaction term in this model), the coefficient on the interaction


term was 5.2 and was strongly significant while the R2 changed very little. Thus, we can

conclude that the market value of investing in computers is substantially higher in high

ORG firms. Investors believe that certain organizational practices make it more valuable,

and vice-versa.

3.3.3. Non-parametric Estimation

The above results suggest that each dollar of computer capital is associated with more

intangible assets in high ORG firms than it is in centralized, low-skill firms. If the stock

market is valuing these firms properly, then this suggests that the benefits of

computerization are likely to go disproportionately to firms that are decentralized.

Figures 3 and 4 graphically capture this idea by plotting results from non-parametric

regressions. Figure 3 is a level plot of fitted values of market value regression on both

computer capital and ORG variables, after netting out effects of other variables. Figure 4

is a contour plot from the same regression. We can see a clear picture of interaction effect

between computers and the ORG variable, which captures most of the decentralized work

practices. Firms which are high in both IT and ORG have much higher market values

than firms which have one without the other. Interestingly, almost all of the effects of IT

and of ORG on market value are concentrated in the quadrant where both asset levels are

simultaneously above the median.


4. Discussion and Conclusions

Our results suggest that the organizational adjustment costs that firms incur when

installing computer capital, including investments in training, organizational change, and

relationship-building, appear to create substantial amounts of intangible assets. The

financial markets treat the organizational assets associated with IT much like other assets

that increase long-term profits but are difficult for competitors to duplicate. By analyzing

several hundred firms over a period of 8 years, the analysis helps to document and

explain the extent to which computerization is associated with both direct and indirect

measures of intangible assets. Furthermore, this approach helps reveal the pattern of

interactions among IT, organizational practices, and market valuations, and thereby

detect complementarities. If these assets are in fact becoming more important in modern

economies, in part because of the information revolution engendered by computers and

communications, then it is incumbent upon us to understand not only particular cases, but

also any broader relationships and patterns that exist in the data.

Our main results are consistent with each of the testable implications about

complementarities between computers and organizational design described in the

introduction:

1) The financial markets put a higher value on firms with more installed computer

capital. The increase in market value associated with each dollar of IT substantially

exceeds the valuation placed on other types of capital.


2) Computer-intensive firms have distinctly different organizational characteristics,

involving teams, broader jobs, and greater decentralization of decision-making.

3) Firms with these organizational characteristics have higher market valuations than

their competitors, even when all their other measured assets are the same.

4) Firms with higher levels of both computer investment and these organizational

characteristics have disproportionately higher market valuations than firms invest

heavily on only one or the other dimension.

Taken together, these results provide evidence that the combination of computers and

organizational structures creates more value than the simple sum of these contributions

separately. The evidence is not consistent with alternative explanations such as

econometric biases created by endogeneity or differences in adjustment speeds of capital

assets. The evidence is consistent, however, with the widespread perception among

managers that information technology is a catalyst for a broad set of organizational

changes (see e.g., Brynjolfsson and Hitt, 2000).

Our interpretation has focused on the assumption that the stock market is approximately

correct in the way it values information technology and other capital investments. The

fact that our results apply to a broad cross-section of the economy over nearly a full

business cycle suggests that fads, industry idiosyncrasies, and investor errors are not

driving the results. Moreover, year-by-year estimation showed a consistently high


valuation of computer capital throughout the 1988-94 period. Our analysis also predates

the large increase in the market value of technology stocks in the late 1990s, and our

sample is disproportionately weighted toward large, established firms rather than new

high-technology entrants; thus, our results are not likely to be sensitive to a “high tech

stock bubble.” Interestingly, productivity analysis by Brynjolfsson and Hitt (1997) found

that the long-run productivity benefits are approximately five times the direct capital cost

of computers, consistent with a valuation of IT on the order of five times higher than the

valuation of ordinary capital.

Finally, what are the implications of our results given the emerging view on the

productivity slowdown after 1973? As Yorukoglu (1996), Greenwood (1997),

Greenwood and Jovanovic (1999), and Bart and Jovanovic (1999) point out, the

productivity slowdown may be explained by the adjustment process accompanying the

transformation from a capital-intensive industrial economy to a computer-intensive

information-based economy. This view is reinforced by R. Hall’s (1999b) interpretation

of investment in reorganization. In his model, when productivity slows down, economic

actors invest more in reorganization, because the search for new production approaches

provides higher marginal benefits than does expanding current production. Our results

broadly support these observations, and further suggest that firms are investing heavily in

reorganization. In particular, decentralized and/or team-based new work practices are

complementary investments to computers, and the overall economic impact of these

complementarities is substantial.


Table 1a. Effects of various assets on firms’ market valuationBaseline Regressions of different models

Market Pooled Fixed EffectWithin

Between

Value OLS w/Year wo/ Year OLSComputer 16.951*** 6.436*** 7.684*** 21.21***

Capital 1.180 0.897 0.837 3.32Physical 0.975*** 1.149*** 1.240*** 0.978***

Capital 0.020 0.054 0.053 0.046Other 0.684*** 0.829*** 0.828*** 0.658***

Assets 0.009 0.012 0.012 0.021

R&D R&D R&D R&DControls Adv Adv Adv Adv

Year*** Year***

Industry*** Firm*** Firm*** Industry***

R-square 0.8698 0.7244 0.7178 0.8875Observations 4592 4592 4592 4592Key: * - p<.1, **- p<.05, *** - p<.01,


Table1b: Effect of various assets on firms’ market valuationBalanced panel only, between and within regressions

Between Regression Fixed Effect WithinRegression

OLS GLS LAD GLS LADComputer 22.285*** 18.540*** 14.824*** 5.584*** 4.308***

Capital 4.193 1.464 3.545 0.921 1.154Physical 0.968*** 1.014*** 0.984*** 1.244*** 1.169***

Capital 0.049 0.016 0.019 0.055 0.113Other 0.654*** 0.656*** 0.652*** 0.811*** 0.814***

Assets 0.024 0.010 0.088 0.015 0.086

Controls R&D R&D*** R&D*** R&D R&DAdv* Adv*** Adv*** Adv Adv***

Industry*** Industry*** Industry*** Year*** Year***

Firm*** Firm***

R-square 0.892 0.869 0.675 0.681 0.836Observations 3312 3312 3312 3312 3312


Table 1c: Leads and Lags between Changes in Market Value and Computer Investments

Computers (0) Simple Instantaneous Market Value (0) Simple InstantaneousCausal Model Causal Model Causal Model Causal Model

Computers (-1) -0.103 -0.104 Market Value (-1) -0.108 0.0090.128 0.127 0.197 0.128

Computers (-2) -0.019 -0.037 Market Value (-2) 0.144 0.0040.172 0.170 0.151 0.050

Computers (-3) 1.120*** 1.123*** Market Value (-3) 0.045 0.0540.231 0.228 0.065 0.070

Market Value (0) 0.0010 Computers (0) 5.960***

0.0006 2.137Market Value (-1) 0.0014 0.0014 Computers (-1) 14.665*** 0.597

0.0010 0.0010 4.722 4.526Market Value (-2) 0.0006 0.0006 Computers (-2) 16.546*** 16.565

0.0018 0.0018 5.864 14.700Market Value (-3) 0.0001 0.0001 Computers (-3) -3.578 -10.701

0.0006 0.0006 16.759 16.209

Controls year*** year*** Controls Year*** year***

Industry*** Industry*** Industry*** Industry***

* All variables except controls are in yearly changes

Table 1d: Long Difference Estimation

DifferencesMarket Value 1 year 2 years 3 years 4 years 5 years 6 years 7 years

Computers 2.949*** 4.644*** 5.523*** 6.539*** 4.041*** 4.746*** 7.806***

1.031 1.332 1.611 1.655 1.710 1.757 2.370

PP&E 0.368*** 0.695*** 0.973*** 1.244*** 1.226*** 1.240*** 1.206***

0.074 0.079 0.083 0.089 0.093 0.099 0.133

Other Assets 0.863*** 0.852*** 0.828*** 0.837*** 0.853*** 0.848*** 0.728***

0.013 0.017 0.019 0.021 0.020 0.020 0.029

Controls year*** year*** year*** year*** year*** Year*

industry*** industry*** industry*** industry*** industry*** industry*** industry***

Observations 2898 2484 2070 1656 1242 828 414R-square 0.66 0.62 0.66 0.71 0.77 0.84 0.86


Table 1e: Year-by-Year Fluctuation of Market Valuation(robust standard errors)

Years 1987-88 1989-90 1991-92 1993-94

Computer 28.435*** 15.966*** 21.082*** 11.965***

Capital 3.962 3.483 3.647 1.665Physical 0.821*** 0.994*** 1.024*** 0.989***

Capital 0.027 0.034 0.048 0.042Other 0.655*** 0.672*** 0.661*** 0.719***

Assets 0.015 0.015 0.022 0.015

R&D*** R&D** R&D R&DControls Adv Adv** Adv Adv

Year Year Year Year**

Industry*** Industry*** Industry*** Industry***

R2 0.907 0.909 0.840 0.887Observations 1090 1089 1182 1259


Table 2a: Correlations between IT measures and organizational structure

Measure(scale in parenthesis)

ITCapital MIPS TOTPC

Structural DecentralizationSelf-Managing Teams (1-5) .17*** .22*** .20***Employee Inv. Grps. (1-5) .07 .08 .08Broad Jobs (1-5) .07 .12** .10*

Individual DecentralizationPace of Work (1-3) .04 .06 .02Method of Work (1-3) .16*** .20*** .15***Composite: 7 Measures^ .12* .14** .16***Individual Control^ .11* .15** .15**

Team IncentivesTeam Building .15*** .19*** .18***Promote for Teamwork .02 .10* .00

Skill AcquisitionTraining (% staff) .14** .15*** .14**Screen for Education (1-5) .16*** .18*** .21***

ORG Composite .24*** .30*** .25***Spearman partial rank order correlations controlling for industry, employment and productionworker occupation. N=300-372, depending on data availability.Key: * - p<.1, ** - p<.05, *** - p<.01

Table 2b: Unrotated Principal Components for ORG Variable Construction

Work Practices

Loading 1st

PrincipalComponent

Loading 2ndPrincipal

ComponentSelf Managing Teams 0.751 0.006Employee Involvement Groups 0.707 0.176Decentralized Pace Decision 0.528 -0.628Decentralized Method Decision 0.572 -0.456Team Building 0.747 0.250Promote for Teamwork 0.401 0.367Screen for Education 0.466 -0.095Training (% Staff Involved) 0.425 0.408

Percent of Variance Explained 34.8% 12.6%


Table 3a: Effect of IT and ORG on market value

Matched SampleBaseline Estimates

Adding ORG variablealone

Pooled Within Between Pooled BetweenComputer 8.866*** 8.684*** 9.776** 8.399*** 9.292*

1.670 1.393 5.073 1.697 5.061ORG 490.7*** 496.0a

130.5 306.7Physical 0.895*** 1.438*** 0.834*** 0.863*** 0.799***

Capital 0.040 0.088 0.103 0.041 0.104Other 0.859*** 0.728*** 0.876*** 0.859*** 0.876***

Assets 0.028 0.046 0.074 0.028 0.074

R&D*** R&D R&D** R&D*** R&D**

Controls Adv Adv** Adv Adv AdvYear*** Year*** NA Year*** NA

Industry*** Firm*** Industry*** Industry*** Industry***

R square 0.8005 0.7197 0.7987 0.8022 0.8460Observations 1707 1707 1707 1707 1707a: p-value = .107


Table 3b: Effect of interaction on market value

Market Value Pooled Pooledw/other

interactions

Between Betweenw/other

interactions

WithinFirm Effect

WithinInteraction

OnlyComputer 2.231 2.760 4.947 4.988 5.818***

2.195 2.282 5.47 5.588 2.001ORG 314.3** 246.5* 261.3 266.1

135.8 146.5 326.1 346.1ORG x 5.907*** 5.401*** 7.493** 7.394a 2.506** 5.152***

Computer 1.344 1.552 3.735 4.594 1.267 0.880ORG x 0.041 -0.007PhysicalCapital

0.0.35 0.092

ORG x -0.035 0.005Other Assets 0.022 0.059

Physical 0.896*** 0.859*** 0.830*** 0.835*** 1.471*** 1.471***

Capital 0.042 0.051 0.105 0.132 0.090 0.090Other 0.813*** 0.809*** 0.801*** 0.796*** 0.699*** 0.705***

Assets 0.030 0.038 0.082 0.101 0.048 0.048

R&D*** R&D*** R&D** R&D R&D R&DControls Adv Adv Adv Adv Adv** Adv**

Year*** Year*** Year*** Year***

Industry**

*Industry*** Industry*** Industry***

R square 0.8045 0.8047 0.8491 0.8491 0.7196 0.7101Observations 1707 1707 1707 1707 1707 1707

a: p-value = .109


Figure 1. Relative size of market valuation: Between estimates

0

5

10

15

20

25

30

Computers Physical Capital Other Assets

Mar

ket

Val

ue

B. OLS

B.GLS

B. LAD

• 95% confidence interval is drawn for computer coefficients.• Confidence intervals for other assets are too small to be shown on this scale.

Figure 2. Relative size of market valuation: Firm effect within estimates

0

1

2

3

4

5

6

7

8

Computers Physical Capital Other Assets

Mar

ket

Val

ue

OLS

GLS

LAD


Figure 3: Market value 3-D plot by organization and ITNon-Parametrically Estimated Fitted Values via a Local Regression Model

Figure 4: Market value contour plot by organization and ITEstimated Fitted Values via a Local Regression Model


Appendix: Data Description

The variables used for this analysis were constructed as follows:

IT Capital. We have a direct measure of the current market value of each firm'scomputer equipment as reported by Computer Intelligence Corp. The market value wasconstructed for each model of computer. Computer Intelligence calculates the currentmarket value, the replacement cost, of computers, using their current market value tableof computer equipment.

Physical Capital. The source of this variable is Standard and Poor’s Compustat AnnualDataset. We considered two options to construct the variable. The first was to constructthe variable from gross book value of physical capital stock, following the method inHall (1990). Gross book value of capital stock [Compustat Item #7 - Property, Plant andEquipment (Total - Gross)] is deflated by the GDP implicit price deflator for fixedinvestment. The deflator can be applied at the calculated average age of the capital stock,based on the three year average of the ratio of total accumulated depreciation [calculatedfrom Compustat item #8 - Property, Plant & Equipment (Total - Net)] to currentdepreciation [Compustat item #14 - Depreciation and Amortization]. The other simplermethod is to use the net physical stock depreciation [calculated from Compustat item #8 -Property, Plant & Equipment (Total - Net)]. According to the productivity literature thefirst method should be used, but to conduct the market value estimation we adopted thesecond approach to ensure consistency with market value and other assets, which aremeasured in current dollars. The dollar value of IT capital (as calculated above) wassubtracted from this result.

Other Assets. The other asset variable was constructed as the total assets [CompustatAnnual Data item #6] minus the physical capital, as constructed above. This itemincludes receivables, inventories, cash, and other accounting assets such as goodwillreported by companies.

R&D Asset Ratio. Constructed from R&D expenses [Compustat annual item #46].Interestingly, this item includes software expenses and amortization of softwareinvestment. R&D stock was constructed using the same rule in Hall (1993a, b). Sheapplied a 15% depreciation rate, so we followed her lead. The final ratio is simply thequotient of the constructed R&D stock and total assets. Fewer than half of firms in oursample reported R&D expenses. The missing values were filled in using the averages ofthe same industry (SIC 4-digits).

Advertising Asset Ratio. We constructed this from advertising expenses [Compustatannual item #45]. Fewer than 20% of our sample of firms reported the item. We appliedthe same rule with R&D assets ratio.


Market Value. Value of common stock at the end of the fiscal year plus preferred stockvalue plus total debt. In Compustat mnemonic code, it is MKVALF + PSTK+DT, whichrepresents total worth of a firm assessed by financial market.

Organization Variable (ORG). We constructed the variable from items from a surveryconducted in 1995 and 1996. The construction procedure using principal componentanalysis is described in the text. This variable captures the degree of new organizationalpractices identified by Osterman (1994), MacDuffie (1995), and Huselid (1994).


Bibliography

Abel, A. B and O. J. Blanchard (1986). “The Present Value of Profits and CyclicalMovements in Investment.” Econometrica, 54(2), pp. 249-274.

Anand, K and H. Mendelson (1995). "Information and Organization for HorizontalMultimarket Coordination." Stanford University Mimeo.

Appelgate, L, J. I. Cash, and D. Q. Mills (1988). "Information Technology andTomorrow's Manager." Harvard Business Review (November-December).

Autor, D. L, F. Katz and A. B. Krueger (1998). "Computing Inequality: Have ComputersChanged the Labor Market?" Quarterly Journal of Economics, v113 n4, November 1998,pp. 1169-1213.

Bart, H and B. Jovanovic (1999). “The Information Technology Revolution and the StockMarket: Preliminary Evidence.” Working Paper, Department of Economics, New YorkUniversity.

Berman, E, J. Bound and Z. Griliches (1994). "Changes in the Demand for Skilled Laborwithin U. S. Manufacturing Industries." Quarterly Journal of Economics, v109 n2, May1994, pp. 367-397.

Bresnahan, T.F. (1997), “Computerization and Wage Dispersion: An AnalyticReinterpretation,” Mimeo, Stanford Department of Economics.

Bresnahan, T. F, Brynjolfsson, E. and L. M. Hitt (1999). "Information Technology,Workplace Organization, and the Demand for Skilled Labor: Firm-Level Evidence."National Bureau of Economic Research (NBER) Working Paper 7136. Cambridge, MA:NBER.

Brynjolfsson, E and L. M. Hitt (1997). "Computers and Productivity Growth: Firm-levelEvidence." MIT Sloan School Working Paper.

Brynjolfsson, E and L. M. Hitt (2000). "Beyond Computation: Information Technology,Organizational Transformation and Business Performance," forthcoming Journal ofEconomic Perspectives, November.

Brynjolfsson, E and A. Renshaw, and M. Van Alstyne (1997). "The Matrix of Change: ATool for Busniness Process Reengineering." Sloan Management Review, Winter.

Brynjolfsson, E and S. Yang (1999). "Intangible Benefits and Costs of ComputerInvestments : Evidence from the Financial Market." Proceedings of the InternationalConference on Information Systems, Atlanta, Georgia, December, 1997. Revised May,1999.


Chan, L. K. C., J. Lakonishok and T. Sougiannis (1999). Stock Market Valuation of theResearch and Development Expenditures, National Bureau of Economic Research(NBER)Working Paper 7223. Cambridge, MA: NBER.

Dos Santos, B. L., Peffers, K. G., and Mauer, D. C. (1993). “The Impact of InformationTechnology Investment Announcements on the Market Value of the Firm.” InformationSystems Research 4(1): 1-23.

Drucker, P. F. (1988). "The Coming of the New Organization." Harvard Business Review(January-February): 45-53.

George, J. F. and J. L. King (1991). "Examining the Computing and CentralizationDebate." Communications of the ACM 34(7): 63-72.

Gormley, J., W. Bluestein, J. Gatoff and H. Chun (1998), “The Runaway Costs of PackagedApplications,” The Forrester Report, Vol. 3, No. 5, Cambridge, MA.

Griliches, Z. (1981). "Market Value, R&D, and Patents." Economic Letters 7: 183-187.

Griliches, Z. and Cockburn, I (1988). “Industry Effects and the Appropriability Measuresin the Stock Market’s Valuation of R&D and Patents,” American Economic Review(Papers and Proceedings), 78(May): 419-23.

Griliches, Z. and Hausman, J. (1986). "Errors in Variable in Panel Data." Journal ofEconometrics 31(1) February 1986, pp. 93-118.

Greenwood, J. (1997). The Third Industrial Rovolution: Technology, Productivity, andIncome Inequality. Washington, D.C., The AEI Press.

Greenwood, J. and B. Jovanovic (1999). "The Information Technology Revolution andthe Stock Market." American Economic Review Papers and Preecedings, 89(2): 116-122.

Jorgenson, D. W. and B. M. Fraumeni (1995). "The Accumulation of Human andNonhuman Capital, 1948-1984." in Jorgenson, Dale W. (ed.), Productivity. Volume 1.Postwar U.S. economic growth. Cambridge and London: MIT Press, 1995, pp. 273-331.

Hall, B. H. (1990). The Manufacturing Sector Master File: 1959-1987. National Bureauof Economic Research Working Paper 3366. Cambridge, MA: NBER.

Hall, B. H. (1993a). "The Stock Market's Valuation of R&D Investment During the1980s." The American Economic Review 84(1): 1-12.

Hall, B. H. (1993b). "Industrial Research during the 1980s: Did the Rate of Return Fall?"Brookings Papers on Economic Activity: Microeconomics 2: 289-343.


Hall, B. H. (1999). "Innovation and Market Value." University of California BerkeleyMimeo.

Hall, R. E. (1999a). The Stock Market and Capital Accumulation, NBER Working Paper7180.

Hall, R. E. (1999b). Reorganization, NBER Working Paper No. 7181.

Hammer, M. (1990). "Reengineering Work: Don't Automate, Obliterate." HarvardBusiness Review(July-August): 104-112.

Hayashi, F. (1982). "Tobin's Marginal q and Average q: Neoclassical Interpretation."Econometrica 50(January): 213-224.

Hayashi, F. and T. Inoue (1991). "The Relation between Firm Growth and Q withMultiple Capital Goods: Theory and Evidence from Panel Data on Japanese Firms."Econometrica 59(3): 731-753.

Hitt, L. and E. Brynjolfsson (1997). "Information Technology and Internal FirmOrganization: An Exploratory Analysis." Journal of Management Information Systems,Fall.

Huselid, M. A. (1995). "The Impact of Human Resource Management Practices onTurnover, Productivity, and Corporate Financial Performance." Academy of ManagementReview 38(3): 635-672.

Ichniowski, C. K., K. Shaw, and G. Prunnushi. (1997). “The Effects of Human ResourceManagement Practices on Productivity.” The American Economic Review (June).

Im, K. S., K. E. Dow, and V. Grover (1998). “A Reexamination of IT Investment and theMarket Value of the Firm: An Event Study Methodology.” University of South CarolinaMimeo.

Jorgenson D. W. and B. M. Fraumeni (1995). "Investment in Education and U.S.Economic Growth," in D. W. Jorgenson (eds.) Productivity. Volume 1. Postwar U.S.Economic Growth. Cambridge and London: MIT Press. p 273-331.

Lev, B. and T. Sougiannis (1996), “The Capitalization, Amortization, and Value-Relevance of R&D.” Journal of Accounting and Economics 21.

MacDuffie, J. P. (1995). “Human Resource Bundles and Manufacturing Performance:Organizational Login and Flexible Production Systems in the World Auto Industry. ”Industrial and Labor Relations Review (January).

Malone, T. W. (1996). Is Empowerment Just a Fad? Sloan Management Review 38(2).


Mendelson, H. and R. Pillai. (1999) Survival of the Smartest. Working Draft.

Montgomery, C. A. and B. Wernerfelt (1988). "Diversification, Ricardian Rents, andTobin's q." Rand Journal of Economics 19(4): 623-32.

Orlikowski, W. J. (1992). "Learning from Notes: Organizational Issues in GroupwareImplementation." Conference on Computer Supported Cooperative Work, Toronto,Canada.

Osterman, P. (1994). "How Common is Workplace Transformation and Who Adopts It?"Industrial and Labor Relations Review 47(2): 173-188.

Sauer, C. and P. W. Yetton (1997). Steps to the Future: Fresh Thinking on theManagment of IT-based Organizational Transformation. San Francisco, California.

Schaller, H. (1990). “A Re-Examination of the Q Theory of Investment Using U.S. FirmData.” Journal of Applied Econometrics, 5(4). October-December, pp. 309-325.

Tobin, James (1969). "A General Equilibrium Approach to Monetary Theory." Journalof Money, Credit and Banking, 1:1 (February): 15-29.

Triplett, J. E. (1989). "Price and Technological Change in a Capital Good: A Survey ofResearch on Computers." in Technology and Capital Formation, D. W. Jorgenson and R.Landau (eds). Cambridge, MA., MIT Press.

Wildasin, D. E. (1984). "The q Theory of Investment with Many Capital Goods." TheAmerican Economic Review 74(1): 203-210.

Wyner, G. and T. Malone, W. (1997). “Online Bureaucracy or Electronic Market: DoesInformation Technology Lead to Decentralization?” Proceedings of the InternationalConference on Information Systems, Atlanta, Georgia.

Yang, S. (1994). The Relationship between IT Investment and Market Value of Firms.Master's Thesis. MIT Sloan School of Management.

Yorokoglue, M. (1996). “The Information Technology Productivity Paradox.”Unpublished Mimeo, Department of Economics, University of Chicago.

Yoshikawa, H. (1980). “On the ‘q’ Theory of Investment.” American Economic Review70(4): 739-743.

Intangible Assets: How the Interaction of Computers and …ebusiness.mit.edu/erik/itqo final-7-00.pdf · Intangible Assets: How the Interaction of Computers and Organizational Structure

Documents